Millions of older adults suffer from depression. A key barrier to patient engagement in application of effective treatments is the difficulty in quantifying depression severity and anticipating future time spent depressed. More refined models of depression risk would give patients better information about their future course, and allow more effective shared decision making. I propose a four-year, mentored research plan to develop and evaluate a depression risk calculator. This tool will be useful for identifying patients at risk for depression, giving them more information about their depression trajectory, identifying other symptoms associated with depression, and enhancing shared decision making between patients and providers. In order to accomplish this end, I will take courses in longitudinal data analysis and qualitative methods, and gain experience in applying shared decision making in primary care. Using three large longitudinal databases of older adults (CHS, HRS, and IMPACT) I will develop algorithms for a depression risk calculator. I will seek input from five patients previously treated for depression about how information from the calculator would have the most benefit for them, and will tailor the outputs accordingly. I will then ascertain the effects of feedback from a depression risk calculator on patient-level outcomes, particularly patient engagement and use of treatments. I will provide feedback from the calculator to 20 patients, and compare their experiences with those of a control group of 20 patients, followed for one year. This research will generate hypotheses and produce preliminary results for larger studies of using risk models to improve treatment for geriatric depression.